You can find datasets from publically available data repositories such as Kaggle, UCI, Amazon, Google, Microsoft, Awesome, and Government Datasets.

you can find these datasets links here and search according to your need (https://towardsdatascience.com/top-sources-for-machine-learning-datasets-bb6d0dc3378b).

I have found some crybaby datasets are following :

https://research.google.com/audioset/dataset/baby_cry_infant_cry.html

https://www.kaggle.com/c/dsplabcup-detecting-baby-sounds

https://paperswithcode.com/dataset/201-people-infant-cry-speech-data-by-mobile

https://zenodo.org/record/5024791#.Y8-Xz3ZBy5c

https://figshare.com/articles/dataset/Infant_Pain_and_Cry_Data_Labels_csv/16910299/1

Thank you for Arpita interest in my question.

But my research problem is a bit specific, concerned with the classification type of babies' cries. For example: discomfort, hungry, sleepy, ..... . Currently "Donate A Cry" dataset quite similar and available for download, but the dataset is not very well labeled

Light Light There are a few approaches you may take to try to collect the infant cry dataset you want for your research. Among the alternatives are:

1. Inquiring directly with the producers of the SPLANN, Baby2020, and Baby Chillanto datasets about acquiring the data. It's conceivable that the writers haven't answered yet, therefore it's worth contacting them again.

2. Looking for open-source datasets about infant screams in places like Kaggle or OpenML.

3. Inquiring with universities or research institutions that do studies on newborn cries to see whether any datasets are accessible for usage.

4. Creating your own dataset by capturing baby cries and classifying them according on the kind of cry. However, getting access to the newborns and getting proper labels may be problematic.

It's important to note that some datasets may have usage restrictions, such as requiring a license or permission to access the data. Check the terms of use or license agreements for any datasets you receive.

Keep in mind that, as you stated, getting datasets on this issue is tough, so you may need to be creative in order to obtain the data you want.

Thank you for Qamar, I will try to create a new dataset. Looks like that's a viable option right now.